Background: There are concerns that school track recommendations that are mostly based on teachers' judgements of students' performance ('judgement-based recommendations') are more biased by students' SES than school track recommendations that are mostly based on standardized test results ('test-based recommendations'). A recent policy reform of the Dutch educational system has provided us the unique opportunity to compare the effects of students' SES on these two types of track recommendations. Aims: The aim of this study was to examine the differences between test-based and judgement-based recommendations regarding the direct and indirect effect of students' SES at student level and school level. Sample: The sample consisted of 8,639 grade 6 students from 105 Dutch primary schools. Methods: Data were analysed using two-level multilevel mediation models. Results: Track recommendations were higher for high-SES students. This was mostly due to differences in students' prior performance. SES also had a small, direct effect on judgement-based, but not on test-based recommendations. The effects were partly situated at school level. Conclusion: Overall, the results indicated that teachers based their track recommendations mostly on students' prior performance without being biased by students' SES.